Side scan sonar image sample amplification and target detection method based on sonar equation

Abstract

The application provides a side scan sonar image sample expansion and target detection method based on a sonar equation, and belongs to the technical field of surveying and mapping engineering, and comprises the following steps: calculating side scan sonar simulation echo intensity based on a seabed simulation environment; generating a target simulation image sample by using the side scan sonar simulation echo intensity, enhancing the target simulation image sample, and obtaining side scan sonar image sample expansion data set; determining a YOLOv5 network as an original target detection model, training the original target detection model based on the side scan sonar image sample expansion data set, and obtaining a target detection model; and inputting a side scan sonar image to be detected into the target detection model to obtain a target detection result. The application simplifies a side scan sonar working model, solves echo intensity by using a sonar equation, thereby obtaining a simulation image, avoids the problem of insufficient sample representation caused by side scan sonar image sample expansion without considering the working principle of the side scan sonar, and improves detection reliability and accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of surveying and mapping engineering, and in particular to a side scan sonar image sample expansion and target detection method based on a sonar equation. BACKGROUND

[0002] As an important marine detection device, the high-resolution image product obtained by the side scan sonar system has important applications in the fields of seabed topography, bottom material and especially underwater target detection. In recent years, with the development of artificial intelligence deep learning technology, compared with traditional side scan sonar image target recognition, the side scan sonar image target detection technology based on deep learning method has shown higher precision and efficiency. However, the performance of deep neural network is closely related to the training data, and the lack of side scan sonar image dataset is the key and difficulty of restricting the side scan sonar image target detection. Therefore, in order to better construct the side scan sonar image detection model, it is very important to expand and enhance the side scan sonar image dataset.

[0003] The commonly used sample enhancement methods mainly include rotation, cutting, flipping and color space transformation, etc. In the field of side scan sonar image sample expansion, there is a sample expansion method of obtaining side scan sonar image by using optical image for style transfer. Although researchers try to consider the characteristics of side scan sonar image as much as possible, such as shadow, etc., due to the great difference between the original optical image and the target form in the side scan sonar image, the effect is not good. Another sample expansion method is to use the generative adversarial network to train the side scan sonar measured image and then generate samples. However, the generated image may deviate from the scene of the side scan sonar image, and subsequent screening is needed, which cannot automatically complete sample expansion. The existing sample expansion problems greatly affect the reliability and accuracy of the side scan sonar image target detection task.

[0004] Therefore, it is necessary to propose a new side scan sonar image sample expansion and target detection method. SUMMARY

[0005] The present application provides a side scan sonar image sample expansion and target detection method based on a sonar equation, which solves the defects in the prior art that the reliability and accuracy of the side scan sonar image target detection are affected due to the lack of image sample data.

[0006] In a first aspect, the present application provides a side scan sonar image sample expansion and target detection method based on a sonar equation, comprising:

[0007] constructing a seabed simulation environment, calculating the side scan sonar simulation echo intensity based on the seabed simulation environment;

[0008] The side scan sonar analog echo intensity is used to generate a target simulation image sample, the target simulation image sample is enhanced, and a side scan sonar image sample expansion data set is obtained;

[0009] The YOLOv5 network is determined as an original target detection model, and the original target detection model is trained based on the side scan sonar image sample expansion data set, so as to obtain a target detection model;

[0010] The side scan sonar image to be detected is input into the target detection model, and a target detection result is obtained.

[0011] According to the side scan sonar image sample expansion and target detection method based on the sonar equation provided by the application, a seabed simulation environment is constructed, including:

[0012] A side scan sonar image simulation model is constructed according to the working process of the side scan sonar;

[0013] In the side scan sonar image simulation model, a seabed terrain is generated by using a preset three-dimensional image tool, an open source three-dimensional model of a target is obtained, and a simulation sonar is determined;

[0014] A sonar equation is determined according to preset sonar experience parameters, and a sound velocity calculation equation is determined according to preset seabed temperature, salinity and depth parameters.

[0015] According to the side scan sonar image sample expansion and target detection method based on the sonar equation provided by the application, a sonar equation is determined according to preset sonar experience parameters, and a sound velocity calculation equation is determined according to preset seabed temperature, salinity and depth parameters, including:

[0016] The sonar equation is obtained from a sound source level parameter, a propagation loss parameter, a target intensity parameter, a noise level parameter, a directivity index and a detection threshold value;

[0017] The sound velocity calculation equation is obtained from preset seabed temperature, preset seabed salinity and preset seabed depth.

[0018] According to the side scan sonar image sample expansion and target detection method based on the sonar equation provided by the application, the simulation echo intensity of the side scan sonar is calculated based on the seabed simulation environment, including:

[0019] The simulation sonar is initialized from the seabed simulation environment, and a simulation sonar motion trajectory is determined;

[0020] According to a preset downward-looking angle and a preset downward-looking angle change step, sound rays are emitted from the simulation sonar motion trajectory in turn, a layered sound velocity profile is obtained from the sound velocity calculation equation, and a sound ray tracking result is calculated according to the layered sound velocity profile;

[0021] The sound ray tracking result is used to obtain a sound ray space vector value, based on the sound ray space vector value, a ray tracing algorithm is used to calculate a sound ray landing point, and the sound ray landing point is accumulated to obtain a sound ray two-way propagation distance and a sound ray lateral deflection distance;

[0022] The sound ray two-way propagation distance and the sound ray lateral deflection distance are used to calculate a sound ray propagation loss, the sound ray propagation loss is substituted into a sonar equation to obtain the side-scan sonar simulated echo intensity.

[0023] According to the side-scan sonar image sample expansion and target detection method based on the sonar equation, the sound ray tracking result is used to obtain a sound ray space vector value, based on the sound ray space vector value, a ray tracing algorithm is used to calculate a sound ray landing point, and the sound ray landing point is accumulated to obtain a sound ray two-way propagation distance and a sound ray lateral deflection distance, including:

[0024] The first layer sound speed, the second layer sound speed, the first layer incident angle, the second layer incident angle, the first layer depth and the second layer depth of the layered sound speed profile are obtained based on the ray tracing algorithm;

[0025] The sound ray two-way propagation distance and the sound ray lateral deflection distance are calculated based on the first layer sound speed, the second layer sound speed, the first layer incident angle, the second layer incident angle, the first layer depth and the second layer depth.

[0026] According to the side-scan sonar image sample expansion and target detection method based on the sonar equation, a target simulation image sample is generated based on the side-scan sonar simulated echo intensity, including:

[0027] A preset mapping resolution is determined based on a preset downward angle change step, and a 0 gray value grid image is established based on the preset mapping resolution;

[0028] The side-scan sonar simulated echo intensity is distributed into each pixel unit of the 0 gray value grid image based on the initial position and the propagation distance of the side-scan sonar simulated echo intensity, and the target simulation image sample is obtained.

[0029] According to the side-scan sonar image sample expansion and target detection method based on the sonar equation, the target simulation image sample is enhanced to obtain side-scan sonar image sample expansion data set, including:

[0030] The target simulation image sample is added with noise by using a preset multiplicative noise variance in a preset image processing library, and the target simulation image sample after being added with noise is labeled by using a preset image labeling tool, and the side-scan sonar image sample expansion data set is obtained.

[0031] In a second aspect, the present application further provides a side scan sonar image sample expansion and target detection system based on a sonar equation, comprising:

[0032] A construction module is configured to construct a seabed simulation environment, calculate side scan sonar simulation echo intensity based on the seabed simulation environment;

[0033] An enhancement module is configured to generate a target simulation image sample by using the side scan sonar simulation echo intensity, enhance the target simulation image sample, and obtain side scan sonar image sample expansion data set;

[0034] A training module is configured to determine a YOLOv5 network as an original target detection model, train the original target detection model based on the side scan sonar image sample expansion data set, and obtain a target detection model;

[0035] A detection module is configured to input a side scan sonar image to be detected into the target detection model, and obtain a target detection result.

[0036] In a third aspect, the present application further provides an electronic device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the method for side scan sonar image sample expansion and target detection based on a sonar equation according to any one of the above aspects when executing the program.

[0037] In a fourth aspect, the present application further provides a non-transitory computer readable storage medium, which stores a computer program executable by a processor, and the computer program implements the method for side scan sonar image sample expansion and target detection based on a sonar equation according to any one of the above aspects when executed by the processor.

[0038] The method for side scan sonar image sample expansion and target detection based on a sonar equation provided by the present application simplifies a side scan sonar working model, solves echo intensity by using a sonar equation, and thus obtains a simulation image, avoids the problem of insufficient sample representation caused by not considering the working principle of a side scan sonar during side scan sonar image sample expansion, and improves detection reliability and accuracy. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0040] Figure 1 is one of the flowcharts of the method for side scan sonar image sample expansion and target detection based on a sonar equation provided by the present application;

[0041] Figure 2 FIG. 2 is a flowchart of a method for side-scan sonar image sample expansion and target detection based on a sonar equation according to an embodiment of the present application;

[0042] Figure 3 FIG. 3 is a structural diagram of a system for side-scan sonar image sample expansion and target detection based on a sonar equation according to an embodiment of the present application;

[0043] Figure 4 FIG. 4 is a structural diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0044] In order to make the objectives, technical solutions and advantages of the present application clearer, the technical solutions of the present application will be described below in detail with reference to the drawings of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative efforts fall within the scope of the present application.

[0045] Figure 1 FIG. 1 is a flowchart of a method for side-scan sonar image sample expansion and target detection based on a sonar equation according to an embodiment of the present application, as shown in FIG. 1, comprising: Figure 1

[0046] Step 100: Constructing a seabed simulation environment, calculating side-scan sonar simulation echo intensity based on the seabed simulation environment;

[0047] Step 200: Generating a target simulation image sample using the side-scan sonar simulation echo intensity, enhancing the target simulation image sample, and obtaining side-scan sonar image sample expansion data set;

[0048] Step 300: Determining a convolutional neural network YOLOv5 network as an original target detection model, training the original target detection model based on the side-scan sonar image sample expansion data set, and obtaining a target detection model;

[0049] Step 400: Inputting a side-scan sonar image to be detected into the target detection model, and obtaining a target detection result.

[0050] Specifically, the embodiment of the present application constructs a side-scan sonar image simulation model based on a sonar equation, builds a three-dimensional simulation environment, sets simulation parameters, performs sonar equation calculation on rays, obtains simulation values including simulated sonar coordinates, ray distances, and simulated echo intensity, performs resampling and normalization according to the obtained simulated echo intensity, and obtains a final side-scan sonar simulation image. The simulation image is input into an existing YOLOv5 network for training to obtain a target detection model. Finally, a side-scan sonar image to be detected is input into the target detection model to obtain a target detection result. The entire process is as shown in FIG. 2.​Figure 2 as shown.

[0051] Firstly, a simulation model is constructed, a commonly used three-dimensional graphics image software is used to simulate a simulation environment, then a simulation echo intensity is calculated based on the constructed simulation model, then a simulation image and an enhanced simulation image are generated, that is, an expanded and enhanced side scan sonar image sample expansion data set is obtained, a YOLOv5 network is trained by using the training data, a target detection model is obtained, finally, a side scan sonar image to be detected is input into the trained target detection model, and a corresponding target detection result is output.

[0052] The simulation image is obtained by simplifying the side scan sonar working model and solving the echo intensity by using the sonar equation, so that the problem of insufficient sample representation caused by the side scan sonar image sample expansion without considering the working principle of the side scan sonar is avoided, and the detection reliability and accuracy are improved.

[0053] Based on the above embodiment, a seabed simulation environment is constructed, including:

[0054] A side scan sonar image simulation model is constructed according to a side scan sonar working process;

[0055] In the side scan sonar image simulation model, a seabed terrain is generated by using a preset three-dimensional image tool, a target open source three-dimensional model is obtained, and a simulation sonar is determined.

[0056] A sonar equation is determined according to preset sonar experience parameters, and a sound velocity calculation equation is determined according to preset seabed temperature, salinity and depth parameters.

[0057] Among them, the sonar equation is determined according to the preset sonar experience parameters, and the sound velocity calculation equation is determined according to the preset seabed temperature, salinity and depth parameters, including:

[0058] The sonar equation is obtained from the sound source level parameter, the propagation loss parameter, the target intensity parameter, the noise level parameter, the directivity index and the detection threshold;

[0059] The sound velocity calculation equation is obtained from the preset seabed temperature, the preset seabed salinity and the preset seabed depth.

[0060] Specifically, the three-dimensional graphics image software Blender is used in the embodiment of the application to process the ship and container targets sunk in the seabed. The Blender is a free and open source three-dimensional graphics image software, which provides a series of animation short film production solutions from modeling, animation, material, rendering, to audio processing, video editing and the like. It also has a variety of user interfaces for convenient use in different work, and has advanced film and television solutions such as green screen matting, camera reverse tracking, mask processing and post node synthesis.

[0061] Firstly, a side-scan sonar image simulation model based on a sonar equation is constructed according to a working process of the side-scan sonar: during forward movement of the sonar, rays are sequentially emitted downward from left to right at a certain time interval, forming discrete fan shapes, the rays have initial energy, and after attenuation in a medium, the rays reach a target, a certain amount of energy is reflected by the target, and the energy returns to the emission position via the original route, and after attenuation in the medium again, the energy reaches a position for receiving a return wave, all simulated return waves obtained by the sonar at a position are sequentially arranged horizontally according to propagation distances from left to right, and return wave data at multiple positions are sequentially arranged vertically according to time sequences, and a simulation image is obtained.

[0062] Secondly, a visual three-dimensional scene is built by using Blender software, and a seabed terrain is generated by using an ANT Landscape plug-in, main function areas of the plug-in mainly include: main parameters of the terrain can be adjusted to make main settings and regenerate a terrain grid, the terrain is smoothed, triangulated, renamed and added with a material; a terrain noise wave can be adjusted to adjust noise wave settings and noise wave types, and the terrain is customized; and a terrain replacement can be adjusted to adjust height, attenuation and levels.

[0063] In a specific implementation, a noise wave type is marble, a noise wave base type is Perlin, an X size is 0.3, a Y size is 3, a noise wave size is 3, a shape is a wave, an offset is a sine wave, a sharp change is soft, a special effect type is a gradient gradient, and an intensity is 1, and a typical sand wave terrain can be obtained; a target open source three-dimensional model is obtained in the Internet, the target model is randomly placed on the terrain by using a python interface provided by Blender, and scaling and rotation parameters are randomly adjusted, after the target model is batch imported into the scene, manual fine adjustment can be performed to avoid a situation that a simulation scene does not match reality, for example, a container target is suspended above the seabed terrain; a sonar is simulated in the software and does not have entity modeling, and only exists as a three-dimensional point on a constraint path, a Bezier curve is newly built at a height H from the seabed as a simulated sonar movement path, a sampling point number n of the sonar on the curve is set, and values of H and n can be set according to a size of the simulation scene, in the embodiment of the present application, H is 20 meters, and n is 1000.

[0064] Thirdly, simulation parameters are set, including sonar equation related parameters and temperature-salinity-depth parameters.

[0065] In the embodiment of the present application, the sonar equation is:

[0066] SL-2TL+TS-NL+DI>DT

[0067] wherein, SL is a sound source level, TL is a propagation loss, TS is a target intensity, NL is a noise level, DI is a directivity index, and DT is a detection threshold.

[0068] The sound source level SL can be set to 220 dB, the noise level NL can be set to 63 dB, the directivity index can be set to 20 dB, and the acceptance threshold can be set to 10 dB, which are all empirical parameters of a general active sonar; the sound speed is initialized according to the sound speed calculation formula, and the sound speed profile is calculated when T = 20℃, S = 34‰, Z = 0, 1, …, H.

[0069] In addition, the sound speed calculation formula adopts the following formula:

[0070] c = 1449.2 + 4.6T - 0.055T 2 + 0.00029T 3 + (1.34 - 0.010T) (S - 35) + 0.016Z

[0071] Wherein, T is temperature, unit is ℃, S is salinity, unit is ‰, Z is depth, unit is m.

[0072] Based on the above embodiment, the side-scan sonar simulation echo intensity is calculated based on the seabed simulation environment, comprising:

[0073] The simulation sonar is initialized by the seabed simulation environment, and the simulation sonar motion trajectory is determined;

[0074] According to the preset downward-looking angle and the preset downward-looking angle change step, the sound ray is emitted from the simulation sonar motion trajectory in turn, the layered sound speed profile is obtained from the sound speed calculation equation, and the sound ray tracking result is calculated according to the layered sound speed profile;

[0075] The sound ray spatial vector value is obtained by using the sound ray tracking result, the sound ray landing point is calculated by using the ray tracing algorithm based on the sound ray spatial vector value, and the sound ray two-way propagation distance and the sound ray lateral deflection distance are obtained by accumulating the sound ray landing point;

[0076] The sound ray propagation loss is calculated based on the sound ray two-way propagation distance and the sound ray lateral deflection distance, the sound ray propagation loss is substituted into the sonar equation, and the side-scan sonar simulation echo intensity is obtained.

[0077] Wherein, the sound ray spatial vector value is obtained by using the sound ray tracking result, the sound ray landing point is calculated by using the ray tracing algorithm based on the sound ray spatial vector value, and the sound ray two-way propagation distance and the sound ray lateral deflection distance are obtained by accumulating the sound ray landing point, comprising:

[0078] The first layer sound speed, the second layer sound speed, the first layer incidence angle, the second layer incidence angle, the first layer depth and the second layer depth of the layered sound speed profile are obtained based on the ray tracing algorithm;

[0079] According to the first layer sound speed, the second layer sound speed, the first layer incidence angle, the second layer incidence angle, the first layer depth and the second layer depth, the sound ray double path propagation distance and the sound ray lateral deflection distance are calculated.

[0080] Specifically, the simulation echo intensity calculation in the embodiment of the present application first initializes the simulation sonar according to the seabed three-dimensional simulation environment obtained in the foregoing embodiment, sets an empty object in the Blender software scene as the simulation sonar, sets the initial position of the simulation sonar at one end of the path, and constrains the animation path of the simulation sonar to be the curve.

[0081] Next, according to the preset downward-looking angle parameter and the preset downward-looking angle change step, the downward-looking angle is set to 140° and the step is 0.01°. The parameter values herein are determined according to the specific simulation requirements. The range of the downward-looking angle is 0°-180°, the step is smaller than the downward-looking angle, and the smaller the step is, the more sound rays there are and the slower the calculation is. The sound ray vectors are sequentially initialized, and the sound ray tracking calculation is sequentially performed according to the layered sound speed profile.

[0082] The sound ray landing point is calculated using the existing ray tracing algorithm according to the spatial vector value of the sound ray in the last layer, and the sound ray double path propagation distance S is accumulated. Finally, the propagation loss is calculated using the propagation distance, and the sonar equation is used to obtain the simulation echo intensity. The propagation loss in the embodiment of the present application is 20log 10 S, which is brought into the sonar equation to obtain the echo intensity value when the sonar equation is greater than the detection threshold DT.

[0083] wherein the sound speed tracking calculates the total propagation distance S and the lateral deflection S x as follows,

[0084]

[0085] S x =R1(cosθ2-cosθ1)

[0086]

[0087] In the formula, θ1 and θ2 are the incidence angles of layers 1 and 2, c1 and c2 are the sound speeds of layers 1 and 2, and z1 and z2 are the depths of layers 1 and 2.

[0088] It should be noted that the above-mentioned layers 1 and 2 refer to the first and second layers of the sound speed profile. The sound speed changes with water depth. In order to facilitate calculation, it is generally considered that the sound speed is constant or has a constant gradient within every interval of water depth, that is, one layer.

[0089] Based on the above-mentioned embodiment, the target simulation image sample is generated by using the side-scan sonar simulation echo intensity, which comprises:

[0090] The preset mapping resolution is determined by using a preset downward viewing angle change step, and a 0 gray value grid image is established based on the preset mapping resolution;

[0091] The side scan sonar simulated echo intensity is distributed into each pixel unit of the 0 gray value grid image based on the initial position and propagation distance of the side scan sonar simulated echo intensity, and the target simulation image sample is obtained.

[0092] Specifically, according to the preset mapping resolution, for example, 0.1 m, a 0 gray value grid image is established, that is, a 0 gray value grid image with a resolution of 0.1 m*0.1 m.

[0093] According to the initial position and propagation distance of each simulated echo intensity, the obtained simulated echo intensity is distributed into the 0 gray value grid image pixel unit.

[0094] Based on the above embodiment, the target simulation image sample is enhanced to obtain a side scan sonar image sample expansion data set, including:

[0095] The preset multiplicative noise variance in the preset image processing library is used to add noise to the target simulation image sample, and the preset image labeling tool is used to label the target simulation image sample after noise addition, to obtain the side scan sonar image sample expansion data set.

[0096] Specifically, after obtaining the target simulation image sample, the simulation image is enhanced according to the preset multiplicative noise variance, and the obtained simulation image is subjected to noise addition processing to obtain the final side scan sonar image sample expansion data set.

[0097] The multiplicative noise function in the skimage package in the Python library is used to add noise to the simulation image with a variance of 0.7, and then the image labeling tool LabelImage is used to label the simulation image, to complete the data set expansion of the side scan sonar image. In specific implementation, a pre-experiment can be performed to determine the value of the multiplicative noise variance.

[0098] The side scan sonar image sample expansion and target detection system based on the sonar equation provided by the present application is described below, and the side scan sonar image sample expansion and target detection system based on the sonar equation described below can be mutually corresponding to the side scan sonar image sample expansion and target detection method described above.

[0099] Figure 3 is a structural schematic diagram of the side scan sonar image sample expansion and target detection system based on the sonar equation provided by the present application, as Figure 3 shown, comprising: a construction module 31, an enhancement module 32, a training module 33 and a detection module 34, wherein:

[0100] The construction module 31 is used to construct a seabed simulation environment and calculate the simulated echo intensity of the side-scan sonar based on the seabed simulation environment; the enhancement module 32 is used to generate target simulation image samples using the simulated echo intensity of the side-scan sonar, enhance the target simulation image samples, and obtain a side-scan sonar image sample augmentation dataset; the training module 33 is used to determine the YOLOv5 network as the original target detection model, train the original target detection model based on the side-scan sonar image sample augmentation dataset, and obtain the target detection model; the detection module 34 is used to input the side-scan sonar image to be detected into the target detection model and obtain the target detection result.

[0101] Figure 4 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 4 As shown, the electronic device may include a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other via the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a side-scan sonar image sample augmentation and target detection method based on sonar equations. The method includes: constructing a seabed simulation environment; calculating the simulated echo intensity of the side-scan sonar based on the seabed simulation environment; generating simulated target image samples using the simulated echo intensity of the side-scan sonar; augmenting the simulated target image samples to obtain a side-scan sonar image sample augmentation dataset; determining a convolutional neural network YOLOv5 network as the original target detection model; training the original target detection model based on the side-scan sonar image sample augmentation dataset to obtain a target detection model; and inputting the side-scan sonar image to be detected into the target detection model to obtain the target detection result.

[0102] In addition, the logic instructions in the memory 430 described above can be implemented in the form of a software functional unit and sold or used as an independent product, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0103] In another aspect, the present application also provides a non-transitory computer readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement a side scan sonar image sample expansion and target detection method based on a sonar equation provided by the above method, the method comprising: constructing a seabed simulation environment, calculating side scan sonar simulation echo intensity based on the seabed simulation environment; generating a target simulation image sample using the side scan sonar simulation echo intensity, enhancing the target simulation image sample to obtain a side scan sonar image sample expansion data set; determining a convolutional neural network YOLOv5 network as an original target detection model, training the original target detection model based on the side scan sonar image sample expansion data set to obtain a target detection model; inputting a side scan sonar image to be detected into the target detection model to obtain a target detection result.

[0104] The device embodiments described above are only schematic, wherein the units described as separate components can or can not be physically separated, and the components displayed as units can or can not be physical units, i.e., they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment. Those skilled in the art can understand and implement without creative labor.

[0105] Those skilled in the art can clearly understand the technical solutions of the various embodiments from the above description of the embodiments, and the various embodiments can be implemented by means of software with the necessary general hardware platforms, and of course, can also be implemented by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part of the prior art that makes a contribution, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0106] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, rather than limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features therein; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A side-scan sonar image sample amplification and target detection method based on a sonar equation, characterized in that, The method comprises the following steps: constructing a seabed simulation environment, calculating side-scan sonar simulation echo intensity based on the seabed simulation environment; generating target simulation image samples using the side-scan sonar simulation echo intensity, enhancing the target simulation image samples, and obtaining a side-scan sonar image sample expansion data set; determining a convolutional neural network YOLOv5 network as an original target detection model, training the original target detection model based on the side-scan sonar image sample expansion data set, and obtaining a target detection model; inputting a side-scan sonar image to be detected into the target detection model to obtain a target detection result; wherein the calculation of the side-scan sonar simulation echo intensity based on the seabed simulation environment comprises: initializing a simulation sonar from the seabed simulation environment, and determining a simulation sonar motion trajectory; according to a preset downward-looking angle and a preset downward-looking angle change step, sequentially emitting sound rays from the simulation sonar motion trajectory, obtaining a layered sound speed profile from a sound speed calculation equation, and calculating a sound ray tracking result based on the layered sound speed profile; obtaining a sound ray spatial vector value using the sound ray tracking result, calculating a sound ray landing point using a ray tracing algorithm based on the sound ray spatial vector value, and accumulating the sound ray landing point to obtain a sound ray two-way propagation distance and a sound ray lateral deflection distance; calculating a sound ray propagation loss based on the sound ray two-way propagation distance and the sound ray lateral deflection distance, substituting the sound ray propagation loss into a sonar equation, and obtaining the side-scan sonar simulation echo intensity; obtaining a sound ray spatial vector value using the sound ray tracking result, calculating a sound ray landing point using a ray tracing algorithm based on the sound ray spatial vector value, and accumulating the sound ray landing point to obtain a sound ray two-way propagation distance and a sound ray lateral deflection distance, comprising: obtaining a first layer sound speed, a second layer sound speed, a first layer incidence angle, a second layer incidence angle, a first layer depth, and a second layer depth of the layered sound speed profile based on the ray tracing algorithm; calculating the sound ray two-way propagation distance and the sound ray lateral deflection distance based on the first layer sound speed, the second layer sound speed, the first layer incidence angle, the second layer incidence angle, the first layer depth, and the second layer depth; generating target simulation image samples using the side-scan sonar simulation echo intensity, comprising: determining a preset mapping resolution using a preset downward-looking angle change step, establishing a 0 gray value grid image based on the preset mapping resolution; based on the initial position and the propagation distance of the side-scan sonar simulation echo intensity, distributing the side-scan sonar simulation echo intensity into each pixel unit of the 0 gray value grid image to obtain the target simulation image samples.

2. The side-scan sonar image sample amplification and target detection method based on a sonar equation according to claim 1, characterized in that, constructing a seabed simulation environment, comprising: constructing a side-scan sonar image simulation model according to a side-scan sonar working process; generating a seabed terrain in the side-scan sonar image simulation model using a preset three-dimensional image tool, obtaining a target open source three-dimensional model, and determining a simulation sonar; determining a sonar equation according to preset sonar empirical parameters, and determining a sound speed calculation equation from preset seabed temperature-salinity-depth parameters.

3. The side-scan sonar image sample augmentation and target detection method based on a sonar equation of claim 2, wherein, determining a sonar equation according to preset sonar empirical parameters, and determining a sound speed calculation equation from preset seabed temperature-salinity-depth parameters, comprising: The sonar equation is obtained from the sound source level parameter, the propagation loss parameter, the target strength parameter, the noise level parameter, the directivity index and the detection threshold; The sound velocity calculation equation is obtained from the preset seabed temperature, the preset seabed salinity and the preset seabed depth.

4. The side-scan sonar image sample amplification and target detection method based on a sonar equation of claim 1, wherein, The target simulation image sample is enhanced to obtain a side scan sonar image sample expansion dataset, including: The target simulation image sample is added with noise by using a preset multiplicative noise variance in a preset image processing library, and the target simulation image sample after the noise addition is labeled by using a preset image labeling tool to obtain the side scan sonar image sample expansion dataset.

5. A side scan sonar image sample augmentation and target detection system based on the equation of sonar, based on the side scan sonar image sample augmentation and target detection method based on the equation of sonar in any one of claims 1 to 4, characterized in that, It includes: The construction module is configured to construct a seabed simulation environment and calculate side scan sonar simulation echo intensity based on the seabed simulation environment; The enhancement module is configured to generate a target simulation image sample by using the side scan sonar simulation echo intensity, enhance the target simulation image sample, and obtain a side scan sonar image sample expansion dataset; The training module is configured to determine a YOLOv5 network as an original target detection model, train the original target detection model based on the side scan sonar image sample expansion dataset, and obtain a target detection model; The detection module is configured to input a side scan sonar image to be detected into the target detection model to obtain a target detection result.

6. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the side scan sonar image sample expansion and target detection method based on the sonar equation according to any one of claims 1 to 4.

7. A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the side scan sonar image sample expansion and target detection method based on the sonar equation according to any one of claims 1 to 4.

Images